Re: [ft-devel] possibility of optimizing for speed with device metrics

[Werner LEMBERG]

> [...] and see that Freetype does all the horizontal advance
> calculations anyway, regardless, then drop them in preference to
> available hdmx data.  (to others: the new flag controls the
> dropping).  Hence the no-difference to speed. So there is a small
> optimization one can make moving that check earlier for normal
> usage, simply trust the hdmx data and not calculating anything, and
> make freetype marginally faster on normal usage.

Sounds reasonable.

> I thought I should mention this on freetype-devel and I'll let
> others put that in as a small evening project or something :-).

Indeed.  Patches welcome :-)


    Werner